The Journal

of Neuroscience,

January

1992,

V(1):

104-l

1.5

4-Aminopyridine-induced Epileptiform Activity and a GABA-mediated Long-lasting Depolarization in the Rat Hippocampus Paul

Perreault

and

Massimo

Avoli

Montreal Neurological Institute and Department of Neurology and Neurosurgery, McGill University, Montreal, Qukbec, Canada H3A 284

Two types of spontaneous field potentials were recorded in rat hippocampal slices after addition of 4-aminopyridine (4AP; 50 @,I). One consisted of brief, epileptiform discharges that occurred at 0.6 f 0.2 set’ in the CA3 and CA1 areas. The other type occurred less frequently (0.036 + 0.013 set?) and was recorded in CAl, CA3, and dentate areas. It corresponded in all regions to an intracellular long-lasting depolarization (LLD; duration, 300-l 200 msec; peak amplitude, 2-15 mV) that was abolished by bicuculline methiodide; therefore, it was mediated by GABA, receptors. Sectioning experiments and the occurrence of propagation failures indicated that LLDs could be initiated by any area of the slice. Furthermore, the propagation of LLDs did not follow any consistent or predictable pattern along known anatomical hippocampal pathways. Finally, neither the occurrence nor the propagation of LLDs was affected when excitatory synaptic transmission was blocked by NMDA and non-NMDA receptor antagonists. In the presence of antagonists of glutamatergic receptors, LLDs disappeared after the omission of Ca*+ or the addition of Cd2+ to the perfusing solution, suggesting that synaptic transmission was required for their generation. These data indicate that 4-AP discloses both interictal epileptiform discharges and LLDs in the rat hippocampus. The first type of activity is presumably related to certain properties of CA3 pyramidal neurons and the neuronal circuit, whereas LLDs originate from the spontaneous, periodic activity of GABAergic interneurons located in any area of the hippocampus, and can propagate to the other areas by the use of nonsynaptic mechanisms. We propose that 4-AP reveals a novel type of interaction among GABAergic interneurons that is based on the accumulation and the dispersion of K+.

Neuroscientistsinterested in the study of ion channelsand synaptic transmissionhave, in recent years, paid considerableattention to aminopyridines, particularly 4-aminopyridine (4-AP) (for review, seeThesleff, 1980). 4-AP is also known to induce Received Dec. 7, 1990; revised Aug. 1, 199 1; accepted Aug. 14, 199 1. This work was supported by grants from the MRC of Canada (MA-8109), Hospital for Sick Children Foundation, and Savoy Foundation to M.A. P.P. was an MRC Fellow; M.A. is an FRSQ Scholar. We are grateful to Ms. Sue Schiller and Mr. Andrew Topaczewski for technical assistance, to Mr. V. Epp for clerical helo. and to Dr. P. Gloor for readine an earlv draft. Correspondence should be addressed to D;. M. Avoli, Montreal Neurological Institute, 3801 University Street, Montreal, Qu6bec, Canada H3A 2B4. Copyright 0 1992 Society for Neuroscience 0270-6474/92/l 20 104-12$05.00/O

seizuresin humansafter poisoning(Spyker et al., 1980)or during therapeutical trials of the drug (Thesleff, 1980). Thus, 4-AP causeselectrographic seizureswhen applied on the neocortex (Szente and Baranyi, 1987) or tonic-clonic seizureswhen injected systemically (Mihaly et al., 1990). The epileptogenic effects of 4-AP have also been studied in the in vitro brain slice. Publishedreports indicate the presenceof interictal epileptiform activity after 4-AP application to CA3 area hippocampal slices (Voskyul and Albus, 1985; Rutecki et al., 1987; Perreault and Avoli, 199l), although ictal dischargesalso occur in immature slices(Chestnut and Swann, 1988; Avoli, 1990). In the CA1 region, three studieshave failed to record any 4-AP-induced epileptiform activity (Buckle and Haas, 1982; Segal, 1987; Perreault and Avoli, 1989), but two others have demonstratedthe presenceof epileptiform field discharges(Voskyul and Albus, 1985;Holsheimerand Lopesda Silva, 1989).The understanding of the effectsinduced by 4-AP on hippocampal slicesis further complicated by reports describing other types of spontaneous potentials. In both CA1 and CA3 areas,Voskyul and Albus (1985) have recorded a type of spontaneousactivity that occurred at a lower rate than the epileptiform dischargesand did not depend on synaptic transmission.We have alsoobserved a synchronous GABAergic potential in these two hippocampal areasduring treatment with 4-AP (Perreault and Avoli, 1989, 1991). A similar GABAergic potential wasrecorded in neostriatal slices(Kita et al., 1985) as well asin rat (Aram et al., 1991) and human neocortical slices(Avoli et al., 1988) treated with 4-AP. Finally, when 4-AP was applied focally onto the CA1 area, spontaneousslow hyperpolarizations occurred at regular intervals (Segal, 1987). In view of the different findings, we investigated the nature of the spontaneousactivity induced by 4-AP in the CA1 , CA3, and dentate areasof the rat hippocampal slice. Some of these results have been reported in abstract form (Perreault and Avoli, 1987). Materials

and

Methods

Experimentswere performedusinghippocampalslicesfrom male Sprague-Dawley rats(200-300pm).Theanimalswereanesthetized with etheranddecapitated. Thehippocampus wasquicklydissected freeand placedin artificialcerebrospinal fluid (ACSF).Transversehippocampal slices(400450 pm thick) werecut on a McIlwain tissuechopperand transferredto a recordingchamberwheretheywerekept at 3435°Con afilter paperat theinterfacebetweenoxygenated ACSFanda humidified atmosphere saturated with 95%0, and5%CO,.Thesliceswereperfused at a constantrate (0.2-0.5ml/min) with oxygenatedACSFof the followingcomposition(in mM): NaCl, 124.0;KCl, 2.0; KH,PO,, 1.25; MgSO,,2.0;CaCl,,2.0;NaHCO,,26.0;andglucose,10.The ACSFwas

The Journal

equilibrated with 95% 0, and 5% CO, to a pH of 7.4. In some experiments, Ca2+ was omitted from the solution or Cd2+ (500 PM) was added in order to block synaptic transmission. The slicing procedure employed in the experiments reported in this article was modified slightly from the protocol used in one of our previous studies (Perreault and Avoli. 1989). Since the 4-AP-induced enileptiform activity recorded in CA1 by ‘some investigators originated from the CA3 area (Voskyul and Albus, 1985) we postulated that the reason others (including ourselves) failed to observe this kind of epileptiform activity in the CA1 region could be related to a low viability of CA3 neurons, which might be injured by the slicing procedure. Therefore, in order to improve the viability of CA3 neurons, special precautions were taken during manipulations of the excised hippocampus to keep the alvear side underneath, facing the wet filter paper. The response properties of the CA 1 neurons of slices obtained with the new protocol were apparently normal as assessed by monitoring the population spike evoked by stimulating the Schaffer collaterals of slices bathed in normal ACSF. However, in the presence of4-AP (50 PM), spontaneous interictal field activity in the CA1 area was seen in 87% of the slices (n = 36) obtained with the modified procedure. It is therefore likely that the overlying white matter of the alveus protects this region when slicing is done with the modified procedure. Drug application.Drugs were applied either to the perfusing ACSF, or focally using a glass micropipette (5-10 pm tip diameter) filled with a known amount of drug dissolved in ACSF. 4-AP was added to the perfusing ACSF to achieve a final concentration of 50 PM. In some experiments, the following drugs were also added: 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX, l-20 PM), 6,6-dinitroquinoxaline-2,3-dione (DNQX; l-10 PM), 3-3 (2-carboxy piperazine-4-yl) propyl-l-phospho(APV, 10-100 nate (CPP, l-5 PM), DL-2-amino-5-phosphonovalerate PM), bicuculline methiodide (BMI; 20-50 PM), theophylline (300 PM), atropine (1 PM), propranolol ( 10 PM), phentolamine ( 10 PM), cimetidine (20 FM), strychnine (1 PM), and naloxone (2 PM). In some cases, BMI (50 PM) was applied focally. All chemicals were acquired from Sigma with the exception of CPP and CNQX, which were obtained from Tocris Neuramine, and phentolamine, which was obtained from Ciba Geigy. Recordingandstimulation.Electrodes for intracellular recordings were filled with 4 M K-acetate (resistance, 40-70 MO) or a mixture of K-acetate (4 M) and 2(tri-methyl-amino)-N-(2,6-dimethylphenyl) acetamide (QX-314, 50 mM; Astra), whereas for extracellular recordings, lower-impedance electrodes (2-10 Ma) were filled with 2 M NaCl. Intraand extracellular signals were fed to high-impedance amplifiers with a bridge circuit that allowed current to pass through the intracellular recording electrode. The bridge was monitored carefully throughout the experiments and adjusted as necessary. Signals were displayed on a Gould chart recorder and oscilloscope and in some cases recorded on tape for later analysis. Intracellular recordings were obtained from pyramidal cells in the CA3b and CA1 b regions as well as from dentate granule cells (Lorente de No, 1934). Extracellular activity was recorded by positioning the electrode in the stratum (s.) pyramidale of the CA1 b, CA3b, or CA3c areas. In the dentate region, extracellular signals were recorded by positioning the electrode in the s. granulosum of the upper blade of the dentate area. Mossy fiber stimuli were delivered through bipolar stainless steel electrodes placed in the granule cell body layer or in the hilar region of the dentate area. Dentate granule cells were activated orthodromically by positioning the stimulating electrode on the subicular side of the hippocampal fissure, along the perforant path, whereas the electrode was placed in the CA1 s. radiatum to stimulate the Schaffer collaterals. Sectioningexperiments.A series of cutting experiments was carried out to establish the origin and the pathway(s) involved in the propagation of spontaneous activity induced by 4-AP in the hippocampal slice. A microknife was used for each cut in the area of interest. After cutting, a stimulating electrode was positioned on either side of the cut pathway to ensure that the section was complete and that the fibers remained functional. To confirm the accuracy of the section, the tissue on either side of the cut was also separated by a distance of 50-500 pm. The Schaffer collaterals were sectioned by cutting through the s. radiatum along an axis perpendicular to the s. pyramidale from s. pyramidale to the edge of the s. lacunosum-moleculare. This cut was made at the level of the CA lc subarea to avoid the presence of any CA2-CA3a neurons on the subicular side of the cut. Sometimes the cut was enlarged from the alveus to the hippocampal fissure to isolate CA1 from CA3. The CA3 area was isolated from the CA1 and the dentate regions by enlarging this cut from the hippocampal fissure to a junction point

of Neuroscience,

January

1992,

12(l)

105

between the dentate gyrus and the hippocampus (part of the CA4 area therefore remained on the dentate side of the cut). Dentate and CA1 regions were separated by sectioning carefully along the fissure to make sure that the s. lacunosum-moleculare remained on the CA1 side. In other slices, the mossy fiber pathway was interrupted by sectioning the slice from the s. oriens to the hippocampal fissure along an axis perpendicular to the s. pyramidale to CA3c area. Timedelay measurements. For these experiments, the extracellular signals were recorded simultaneously in all areas of the slice and displayed on a digital oscilloscope. For each trace, the onset of the 4-APinduced spontaneous potentials was determined as the time of the earliest deflection of the baseline recording. All quantitative results are expressed as mean -C SD. The results obtained were compared by using the Student’s t test and considered significantly different if p < 0.05.

Results Spontaneoussynchronousactivity induced by 4-AP Bath application of 4-AP to hippocampal slicesproduced the appearanceof two types of spontaneousfield potential that could be discriminated by their shapeand frequency of occurrence (Fig. 1B). One type was recorded simultaneously in the CA3 and CA1 areasand consistedof brief interictal-like epileptiform field potentials occurring at a frequency of 0.3-1.3 set* (0.6 f 0.2 seem’;n = 31 slices). In both regions, these spontaneous epileptiform discharges,when recorded in the s. pyramidale, consisted of 20-70 msec positive waves with negative-going population spikes(Fig. 1C). However, asshown in Figure 1, Ca and Cb, the epileptiform events recorded in CA3 were more robust, asindicated by the greater number of population spikes (4-l 1 in CA3 as compared to 2-5 in CAl) and their higher amplitude (2 + 1.2 mV, n = 4 slices;range, 0.54 mV in CA3) as compared to CA1 bursts (0.3 f 0.5 mV; n = 10 slices;p < 0.05; range, 0.05-1.3 mV). The epileptiform activity in CA1 was seenin 87% of the slicesanalyzed (n = 36), and in 19.3% of thesecasesconsistedof a simplepositive-going wave without any visible population spikes.Epileptiform activity in CA3 was more prominent in the CA3c subfield and was never observed in the dentate area (n = 60 slices). The other type of spontaneouspotential occurred at a lower rate (0.036 f 0.013 see-I; n = 23 slices;range,0.005-0.07 set?) and lastedlonger(400-l 000 msec)than the brief interictal events. It wasrecorded simultaneously,with small delays, in all regions of the hippocampal slice (Fig. 1B). Furthermore, as shown in Figure l& there was a significant refractory effect of this potential on the interictal activity as indicated by the 1.9-6.4~ longer interdischarge interval following the occurrence of this potential (2.9 + 1.7 x longer; n = 6 slices).When recorded in the cell body layer of CA1 or dentate gyrus, the secondtype of activity appeared

as a smooth, positive-going

potential,

whereas

in the pyramidal layer of CA3 there was usually a faster, negative-going deflection precedinga smootherand low-amplitude positive-going wave. At times, this second type of potential could be precededby an epileptiform event that, however, was of larger amplitude in the CA1 subfield than in CA3 (seeexpandedtracesin Fig. 1Cc, Cd, aswell asin the top panel in Fig. 6A). Other times the long-lasting spontaneouspotential was precededby a brief positive wave reminiscent of a field EPSP (seeextracellular recording traces in Figs. 2A, 3A). Eflects of 4-AP on CA3 neurons Simultaneousintra- and extracellular recordings revealed that the interictal epileptiform field dischargesin the s. pyramidale of the CA3 region correspondedto brief depolarizations of the neuronal membranewith overriding fast action potentials (Fig.

106

Perreault

Dentate

and Avoli

l

4-AP

and

Rat Hippocampus

e 12mV

C ‘A10.4mV “-?1

L-Mr---‘0.4mV

Figure 1. Spontaneous field activity induced by 4-AP. A, Diagram of the hippocampus showing the position of the recording electrodes in CAl, CA3, and dentate areas where the signals shown in B and C were recorded. B, Simultaneous field recordings showing the two types of spontaneous activity produced by 4-AP in CAI, CA3, and dentate regions

(Dentate).Interictal burstsin CA1 and CA3 areasoccurredat regular intervals,whereasa lessfrequent,slowerpositivewave wasrecorded in all regions.C, a andb showoneexampleof interictalburstsoccurring in CA1 andCA3, respectively,anddisplayedat a shortertime basefor clarity (seelabelingin B); c-e areexpandedtracesthat showthe onset of the slowpotentialthat occurredin CA 1, CA3, and dentateareas, respectively(seelabelingin B).

2A). This intracellular activity was reminiscent of the paroxysmal depolarization shift produced by severalconvulsant drugs (for review, seeAlger, 1984). The secondtype of field potential disclosedby 4-AP coincided with a 2-12 mV (8.8 + 2.6 mV; n = 10 cells) depolarization of the neuron’s membrane that lasted 350-1200 msec (548 f 230 msec; n = 14 cells)and was followed by a hyperpolarization (amplitude: 6.5 + 1.8 mV; range, S-10 mV, n = 10 cells; duration: 1440 * 598; range, 500-2000 msec; n = 10 cells). It resembled the long-lasting depolarization (LLD) previously reported by us in CA1 pyramidal cells, where it is caused by the activation of GABA, receptors located on the dendrites of pyramidal cells (Perreault and Avoli, 1989). This wasalso confirmed in the CA3 subfield by applying BMI focally onto the apical dendritic area of pyramidal cells. As shown in Figure 2B, BMI reversibly blocked the slow depolarization, thereby uncovering a hyperpolarizing potential (n = 4). As in CAl, the onset of the LLDs recorded intra- or extracellularly in CA3 displayed great variability. However, in most casesthey were preceded by a burst of action potentials (Fig. 2C, top trace) or a small depolarization that correspondedpresumably to an EPSP (Fig. 2A, top trace). In -5% of the cases

(e.g., Fig. 2C, bottom trace), LLDs were initiated by a brief hyperpolarization of the membrane, presumably an IPSP. In the latter case,action potentials occurred near the peak of the IPSP and on the initial part of the rising phase of the LLD. These action potentials were recorded in more than 50% of the CA3 neurons (n = 7) and had features similar to the action potentials associatedwith the LLDs recorded in CA 1, including varying amplitudes, IS-SD fractionation, and insensitivity to membrane hyperpolarization (Perreault and Avoli, 1989). A smalland fast-risingbriefdepolarization reminiscentof an EPSP was sometimesobserved near the LLD onset (Fig. 2C, arrow on top trace). Stimulation of the mossy fibers evoked a burst followed by an LLD (Fig. 20). However, when the stimulation was applied too frequently or shortly after the occurrence of a spontaneous LLD (< 2-5 set), an additional LLD could not be evoked. Stimulations applied at slightly longer intervals (> 3-l 5 set) evoked LLDs of lower amplitude and duration. LLDs were therefore followed by an absolute and a relative refractory period. This property was not sharedby the epileptiform bursts that could follow afferent fiber stimulations given at short intervals (< 1 set). LLDs were also evoked in CA3 neurons by stimulating anywhere in the slice. Efects of 4-AP on CA1 neurons The effectsinducedby 4-AP on CA 1 neuronshave beenreported in a previous publication (Perreault and Avoli, 1989).However, the modified slicingprocedure usedfor the presentexperiments (seeMaterials and Methods) prompted us to reinvestigate the effects induced by 4-AP in this region. As shown in Figure 3A, the brief interictal events recorded extracellularly in the CA 1b area (bottom trace) were correlated intracellularly (top trace) with EPSPs (1.5-11 mV) that lasted up to 80 msec, and/or EPSP-IPSP sequences.Field dischargeswere correlated with intracellular bursts of action potentials in only 1 out of 24 neurons impaled from 24 different slices.In addition, LLDs similar to those observed in CA3 were also observed in the CA1 area. Comparedto our previously publisheddata (Perreault and Avoli, 1989), one difference here wasthat in 72% of the slices(n = 18) many spontaneousLLDs (> 50%) were triggered by a field epileptiform discharge.This finding wasprobably related to the improved viability of CA3 neuronsin the experimentsreported here (seeMaterials and Methods). However, in agreementwith our previous study, in 80% of the slices(n = 12), the response to orthodromic stimulations consisted of a single population spike. Even with high-intensity stimuli, additional population spikesor burstsfailed to occur (Fig. 3B). However, a small burst similar to the spontaneouslyoccurring potential could be inducedby stimulating the Schaffercollateralswith a low-intensity (lo-50 hA), brief train (three to five stimuli at 150-250 Hz) that mimicked presumably the excitatory input coming from CA3 (Fig. 3Cb). The CA3 origin of CA1 burstswas also suggestedby the fact that CA3 burstsprecededthosein CA1 by 8-24 msecdepending on electrode location (n = 5 slices;Fig. 30, left). This latency correspondedto a propagation velocity of 0.2-0.4 m/set, which is within the expected range for activity propagating along the Schaffer collaterals (Andersen et al., 1978). This interpretation was confirmed in experiments where a cut was made between CA1 and CA3 (n = 8 slices).The sectioning of the Schaffer collaterals abolishedin CA1 both the interictal dischargesand the dischargesthat preceded spontaneousLLDs (seeFigs. 30,

The Journal

-CJlm”

. I

c

of Neuroscience,

200 ms

J--\_II

D

--.I

10 mV

IOOms

-J-

20 mV -I 100 ms

A\ A

1992,

12(l)

107

Figure 2. Spontaneous bursts and LLDs produced by 4-AP in CA3 neurons. A, Simultaneous intracellular (top truce) and extracellular (bottom truce) recordings obtained from the s. pyramidale of the CA3c area. Intracellular bursts were correlated with the interictal field events, whereas an LLD (asterisk on top truce) corresponded to the other multiphasic event. Arrows point to two low-amplitude, spontaneous PSPs. B, Focal application of BMI (50 PM) onto the dendrites of a CA3 neuron blocked the spontaneously occurring LLD, indicating that it was mediated by GABA, receptors. C, Two examples of spontaneous LLDs taken from the same neuron showing the variability in LLD onset. In the top truce, the LLD was initiated by a burst followed by an EPSP (arrow), whereas the LLD shown in the bottom truce started as an IPSP associated with low-amplitude action potentials. D, Mossy fiber (triangle) stimulations evoked responses similar to those that occurred spontaneously (same neuron as in C). Action potentials are cut in A and B. Calibrations in D apply to C. Resting membrane potential of the cell in A was not available; B-D, -70 mV.

BMI

--I’d-II

January

J--

I 5mV -

10.5 mV

500 ms

C

B#----a

400 pi

I

J

5mV

CA+!--

IO.5 mV

20 ms

D&sJ CAIA

b

20 ms

m --+---J~.O

mv 20 ms

Figure 3. Spontaneous bursts and LLDs produced by 4-AP in CA1 pyramidal neurons. A, Simultaneous intracellular (top truce) and extracellular (bottom truce) recordings obtained from the s. pyramidale of CAlb. Brief extracellular interictal events were correlated with short-lasting depolarizations (presumably EPSPs), whereas an LLD (asterisk on top truce) occurred during the slow field potential. Arrows point to two low-amplitude, spontaneous PSPs. B, The extracellular response obtained in the presence of 4-AP in the CA1 s. pyramidale after high-intensity (400 PA) stimulation of the Schaffer collaterals consisted of a single population spike. Cu, In the same slice as in B, a brief train of low-intensity stimulation of the Schaffer collaterals (30 PA) evoked a small burst (a) that was very similar to the one that occurred spontaneously (b). D, The traces on the Zeji show interictal bursts recorded simultaneously in the s. pyramidale of CA1 b and CA3c on an intact slice. Following a cut from the alveus to the fissure, this activity disappeared in CA1 but remained in CA3. The action potential riding on the EPSP at the onset of the LLD in A was cut. Resting membrane potential of the neuron in A was -60 mV.

108

Perreault

and Avoli

* 4-AP

and

Rat Hippocampus

A Intra.

-------/A

bv

Extra.

Anti.

1 Spont.

It -I

40 mV

40 ms

potentials (Fig. 4B, Ortho.). In 44% of the cases(n = 72, eight cells), action potentials of varying amplitudes occurred during spontaneousor evoked LLDs and displayed features identical to those describedfor the CA3 and CA1 regions(insetsin Fig. 4B). 4-AP and spontaneoussynaptic potentials In addition to the two types of synchronous activity described above, 4-AP also induced the appearance of low-amplitude spontaneouspostsynaptic potentials (PSPs)that increasedthe baselinenoise of the intracellular recordings (seearrows in the intracellular traces in Figs. 2A, 3A). However, the nature and expressionof this spontaneoussynaptic activity differed from one region to another. The traces in Figure 5 show typical examplesof PSPselicited by 4-AP in neuronsof the CA3, CAl, and dentate areas.These recordings indicate that this activity had greater amplitude in CA3 cellsthan in CA1 neurons. Furthermore, thesePSPswere virtually absentin dentate neurons. This difference is alsoillustrated in the histogramsof Figure 5, which show the frequency distribution of PSPsof different amplitude in 10 neuronsimpaled in the CA3, CA 1, and the dentate regions.These resultsdemonstrate, therefore, a correlation between the propensity of a given area to produce epileptiform dischargesand the occurrenceand amplitude of PSPs,the order being CA3 > CA1 > dentate.

120mV

t 1s

Figure 4. Spontaneous andevokedLLDs evokedby 4-AP in dentate granulecells.A, Simultaneous intracellular(Zntru.) and extracellular (Extra.) recordingsobtainedfrom the s.granulosum of thedentatearea showingthe spontaneous occurrence of a doubletof LLDs. B, Examples of responses obtainedin anotherdentategranuleneuronafter stimulation (triangles)of the mossyfibers(Anti.) andthe perforantpath(Orthe.). Note the similaritieswith the spontaneously occurringpotential of late actionpotentialsin each (Spent.).The insets showthe presence case. Action potentials are cut in the traces shown in B at slow time base. Resting membrane potential of the neuron in A was -6 1 mV, in B, not available.

6). Therefore the epileptiform activity produced by 4-AP in the CA1 area originates in CA3. Effects of 4-AP on dentate neurons Simultaneousintra- and extracellular field recordings obtained from the s. granulosumrevealed that the spontaneous,positivegoing deflection on the extracellular record correspondedto a brief hyperpolarization followed by an LLD lasting 400-900 msec (Fig. 4A). The peak amplitude of the LLD measured515 mV (n = 9) and was correlated with the decaying phaseof the slow positive wave on the extracellular record, whereasthe hyperpolarization was better correlated with the brief positive wave of the extracellular record. In about 5% of the slices(n = 60) LLDs appearedin doublets (Fig. 4A) or were followed by a seriesof 3-l 5 decrementing depolarizations separatedby an = 1 set interval. LLDs recorded in dentate neurons were initiated by an IPSP and were followed by a 0.9-5.0 set hyperpolarization (n = 8 cells). The GABAergic nature of the LLD seen in this region wasconfirmed by BMI application (20 PM), which abolished the early IPSP and the LLD (n = 2; not illustrated). Similar LLDs were evoked by stimulating the mossyfibers(Fig. 4B, Anti.) or any other part of the slice. Perforant path stimulations evoked an EPSP that preceded the entire sequenceof

Origin and propagation of LLDs To understand better the origin of the LLDs, we performed a seriesof sectioning experiments in which fiber pathways were severedand someareasof the sliceswere isolated. One typical example of the results obtained in 12 slicesis shown in Figure 6A. The top panel showsthe two types of spontaneousactivity produced by 4-AP in the intact slice where field potentials were recorded simultaneouslyin CA3, CA 1,and dentategyrus. When the slicewascut betweenCA1 and CA3 (Fig. 6A, secondpanel), all epileptiform activity was abolished in the CA1 area but remainedpresentin CA3. However, spontaneousLLDs continued to appear simultaneously, with brief delay, in all areasof the slice after this procedure. SpontaneousLLDs and bursts persistedin the CA3 areawhen this region was completely isolated from the remainder of the slice (third panel). Under these conditions, LLDs were still recorded nearly simultaneouslyin both the CA1 and the dentate regions.Further dissectionalong the fissure to isolate completely CA1 and the dentate gyrus demonstrated that LLDs could be generatedindependently by each isolated area (Fig. 6A, bottom panel). The effectsof these cuts on the frequency of occurrence of the LLDs recorded in CA3, CA 1, and dentate regionsare showngraphically in Figure 6B. Compared to the values obtained from intact slices,the LLD rate in CA3 wasnot significantly changedafter all sections had been studied. However, the LLD rate in CA1 and dentate was depressedafter the isolation of the CA3 region and decreasedfurther in both regionsafter they had beenisolatedfrom each other (p < 0.01). In four other slices,we found that the mossy fibers were not required for LLD propagation between CA3 and the dentate gyrus becausethis activity could spread in both areasafter sectioningthis pathway (not shown). These resultstherefore failed to identify any specificpacemakerregion or any obligatory LLD propagationpathway and raisedthe possibility that this potential could be initiated in any of the hippocampal regions.This wasconfirmed in experiments in which we observed propagation failures. While recording simulta-

The Journal

of Neuroscience,

January

1992,

72(l)

109

2.5

1 .o

i 6

CA1

-J

C

7

0.5

8 ;

0 2.5

2

2.0

75

1.5

h 22

1.0 0.5

o0 LL t?! L 2.0r

Dentate

-

10.4 mV --------A*

tIlV 1s 0

43

2.5 mV T -8

-4

0

4

8

12

18

q n

CA3 CAi Dentate

20

Amplitude of PSPs (mV) Figure 5. Spontaneous synaptic potentials recorded intracellularly in CA3 and CA 1 pyramidal cells and dentate granule cells in the presence of 4-AP. The traces on the left show the small spontaneous potentials that occurred in one neuron of the CA3c (A), CA 1b (B), and dentate areas(C). The corresponding histograms on the right show the frequency distribution of spontaneous synaptic potentials of different amplitudes recorded in 10 different neurons impaled in each region and taken from 10 different slices. For each cell, the frequency distribution of the synaptic potentials was calculated for a period of 100 sec. Rest V, of the neuroninAwas-61mV,B, -67mV,andC,-59mV.

neously in CA1 and CA3, we observed in five slicesLLDs in CA3 that failed to propagateto CAl, and in six other sliceswe noted LLDs occurring in CA1 without any correspondingpotential in CA3 (not shown). The origin of LLDs wasfurther investigated in eight slicesby measuring the delay between the onset of the LLDs recorded simultaneously in the cell body layer of CA3, CA 1, and dentate areas. These results confirmed that there was no singlearea of origin for this type of activity. In 39.6% of the cases(n = 56) spontaneousLLDs were recorded first in CA3, while 3 1% of them started in CA1 and 2% in the dentate area. Moreover,

Figure 6. Effects of sections of fiber pathways on the propagation of bursts and LLDs. A, The top panel shows the spontaneous LLDs and bursts recorded simultaneously in the s. pyramidale of CA3c (top trace) and CAlb areas (middle trace) and in the s. granulosum of the dentate region (bottom trace) in an intact slice exposed to 4-AP. LLDs remained coupled in all regions after sectioning the fibers between CA1 and CA3 (second panel) and appeared simultaneously in CA1 and dentate after the CA3 area had been severed from the rest of the slice (third panel). LLDs could be generated by each isolated region of the hippocampus (bottom panel). B, Histogram displaying the effects of these cuts on the frequency of occurrence of the LLDs recorded in each area.

consistedof LLDs originating from CA3 and invading both CA1 and dentate areas simultaneously, with no measurabledelay between the two. In this situation, however, compared to the intact slice,the time delay increasedbetweenCA3-dentate (from 14.3 f 14 msec,y1= 54 events, in control to 51.7 f 21.7 msec,

about 17% of the spontaneous LLDs appeared first, with no measurable delay, in both CA3 and dentate or in both CA1 and

it = 24, after the cut; p < 0.01) and CA3-CA1 (from 14.3 * 17.4, n = 52 events, to 65.4 f 26.6 msec, n = 19; p < 0.01).

dentate. Finally, in 10.4% of the cases,LLDs appeared in all

Less often, LLDs were initiated in CA1 (21% of cases)and dentate (8% of cases)regions.LLDs recorded simultaneouslyin CA1 and dentate gyrus after complete separation from CA3 were initiated first in dentate gyrus in 90%of the casesanalyzed, with time delays ranging from 50 to 207 msec(101.7 f 57.3 msec; y1= 11 events; p < 0.01 compared to intact slice). Stim-

three areas with no measurable

delay between

any of the three

regions.Delays of propagation could be underestimatedin these experiments if spontaneouspotentials originated somewherebetween the three recorded sites and propagated simultaneously toward two or all of them. This may have been the casewhen no delay was observed between regions. In any given slice, LLDs invaded the different areasof the slice by following several patterns. For instance, although the activity originating in CA1 usually propagatedfirst to CA3 and then to dentate,

in the case of seven events the dentate area was

invaded before CA3, while five other LLDs invaded simultaneously CA3 and dentate areas.When sliceswere cut between CA 1 and CA3 from the alveus to the fissure,the most frequently found pattern

of propagation

(62% of the cases; n = 14 events)

ulation

of the s. radiatum

in CA1 evoked

an LLD

in dentate

gyrus after a similar latency of 50-l 15 msec(88.8 -t 19.0 msec; n = 13). These values would correspond to a propagation velocity of 0.004-0.0 17 m/set, assumingdirect propagation from one recording point to the other. The difficulty estimating LLD velocity was compounded by the uncertainty of both the delay measurements(LLD origin could not be determined precisely) and the distance values (propagation pathways not clearly identified by the sectioning

110

Perreault

and

Avoli

- 4-AP

and

Rat Hippocampus

Control

I

Control

Figure 7. Excitatory amino acid antagonists fail to prevent the occurrence and the propagation of LLDs. A, Control traces (top) show a low-speed chart record of the spontaneous activity recorded extracellularly in the s. pyramidale of CA1 b and CA3c and the s. granulosum of the dentate area. After the addition of 10 PM DNQX (middle pane& or 10 PM DNQX and 5 PM CPP (bottom panel), LLDs were still recorded nearly simultaneously in all three regions. B, Plot of population spike amplitude measured in the CA 1 s. pyramidale versus stimulation intensity of the Schaffer collaterals obtained in control condition (open circles), and after the addition of different doses of CNQX. C, Relationship between the frequency of occurrence of the LLDs and the concentration ofCNQX added to the ACSF. These results were taken from the same slice as in B.

DNQX

50 pM)

50

75

100

125

150

175

200

Stimulus Intensity (PA)

n

DNQX + CPP -J-v--

11m”

I ‘10.2

experiments). For comparison purposes, however, we computed the propagation velocity of LLDs spreadingfrom CA3 to CA 1,

assumingthey were conducted directly from one area to the other, and thus we excluded all caseswhere no delay was observed. Under these conditions, the mean propagation delay measured corresponded to a velocity of 0.26 + 0.21 m/set (range,0. l-l .25 m/set; n = 38), which is closeto the lower limit of what would be expected for conducting activity along the Schaffercollaterals(Andersenet al., 1978;Holsheimerand Lopes da Silva, 1989). analysis of spontaneous LLDs After the addition of DNQX or CNQX (3-10 PM), which are antagonists of cu-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA)/kainate receptors(Honor6 et al., 1988),the epileptiform activity in CA3 and CA1 was abolishedbut LLDs continued to occur simultaneously in all three regions(n = 10 slices;Fig. 7A, middle panel). The occurrenceof LLDs wasalso not affected when NMDA receptor antagonistsAPV (20-100 PM) or CPP (l-5 MM) were added to the perfusing medium that already contained CNQX or DNQX (n = 16 slices; Fig. 7A, bottom panel). As both NMDA and non-NMDA receptor antagonistsshould completely block synaptic transmissionsat synapsesthat utilize excitatory amino acids as neurotransmitters (Neumann et al., 1988; Davies and Collingridge, 1989; Lambert et al., 1989), these results indicated that LLD propagation in the hippocampusdid not require excitatory amino acid-dependent synaptic transmission.This was confirmed by monitoring simultaneously the occurrence of spontaneousLLDs and the amplitude of the population spikerecorded in the CA 1 s.pyramidale in response to Schaffer collateral stimulation. When excitatory synaptic transmissionin CA 1wascompletely blocked by high concentrations of CNQX (15-20 PM; Fig. 7B), the LLD occurrencerate in the sliceremainedunchanged(Fig. 7C). However, though LLD occurrence was still coupled in all regions Pharmacological

25

ci

-7

4

(4AP 0

14 -

5s

mV

’

o.ooL

’

;

,b

,I5

do

CNQX (PM)

after CNQX application, the propagation delay increasedsignificantly. In five slices,the mean delay between CA 1 and CA3 increased,from a control value of 13.2 f 10.9 msec(n = 38 LLDs) to 59.8 f 25.5 msec(range, 15-125 msec;IZ= 30 LLDs; p < 0.01) in CNQX. The latter was very similar to the value obtained after cutting the Schaffer collaterals and would correspondto an apparent propagation velocity of 0.02-0.2 m/set. As illustrated in Figure 8, CNQX and CPP altered the shape of the LLDs by blocking the epileptiform dischargethat preceded the LLDs in both CA1 and CA3 (n = 10 experiments). The sharp negative deflection at LLD onset in CA3 was also reduced under these conditions, such that LLD shapein CA1 and CA3 becamealmost identical (Fig. 8A). The intracellular correlate of these effects is demonstrated in Figure 8B, which illustrates LLDs recorded in a CA3 neuron impaled with a QX3 14-filled microelectrodeto prevent action potential generation: although CPP (2.5 MM) did not produce any significant effect on the spontaneousLLD, CNQX (3 or 9 PM) completely blocked the initial high-amplitude EPSP (n = 10 cells). Spontaneous LLDs recorded from both CA1 and CA3 pyramidal cells in the presenceof CNQX were initiated by an early IPSP and, in = 20% of the cases,were followed by a late hyperpolarization. The occurrence of variable-amplitude action potentials associated with the LLDs was not affected by excitatory amino acid antagonists.Theseresultstherefore suggestthat, even though fiber pathways utilizing non-NMDA receptorsmay participate in the generation of LLDs, they are by no means necessary.As no EPSP could be detected in those experiments, not even by hyperpolarizing the neuron membrane,it is likely, rather, that LLD generationwas involving only the activity of GABAergic interneurons. The mechanismsunderlying the propagation of LLDs from one hippocampal areato another were further studied by omitting Ca*+from the ACSF (n = 5) and by adding Cd2+(n = 1) to the medium in order to block CaZ+entry through voltage-sen-

The Journal

of Neuroscience,

January

1992,

12(l)

111

Figure 8. LLDs recordedintracellu-

CNQX

+ CPP

CNQX

Wash

1

500 ms

sitive Ca*+ channels. The top panel in Figure 9A showsspontaneousLLDs recorded simultaneouslyin CA1 and CA3 during perfusion with ACSF that contained 4-AP aswell asCNQX (20 KM) and CPP (5 PM). When Ca2+wasomitted from the perfusing solution, LLDs were completely blocked (Fig. 9A, middle panel) and reappearedafter Ca2+reintroduction (Fig. 9A, bottom panel), indicating that synaptic transmissionwasrequired for LLD generation. Moreover, when GABAergic synapseswere blocked by adding BMI (30-60 PM) to the ACSF already containing 4-AP, CNQX, and CPP, all activity was suppressed(Fig. 9B; n = 5). This finding implies that the only synchronousactivity occurring when excitatory amino acid receptorsareblockedwasthe GABAmediated LLDs recorded in pyramidal neurons and granule cells.In two slices,however, a spontaneouslow-amplitude, negative-going potential (when recorded at the cell body layer) remained after BMI application (not shown). The shapeof this potential was reminiscent of the late phaseof the control LLD and was therefore presumably mediated through GABA, receptors that are not sensitive to BMI (Dutar and Nicoll, 1988). If synaptic transmission were participating in the coupling of intemeurons in the different areasof the slice,thesedata would suggestthat it also involved GABA, receptors. The involvement of some other putative neurotransmitters such asadenosine,ACh, noradrenaline, histamine, glycine, and opiates, whose effects have been previously identified in the hippocampus, was also investigated by the use of specific antagonists. As summarized in Figure SC, none of theseantagonists produced significant changesin LLD frequency, nor, moreover, was the propagation of LLDs affected in the presenceof these antagonists(not shown). Discussion

4-AP and epileptiform activity We have found that 4-AP induces spontaneousinterictal-like, epileptiform activity that originates in CA3 and propagatesto

300 ms

larly and extracellularly after the addition of excitatory amino acid antagonists.A, LLDs recordedsimultaneouslyin the s. pyramidaleof CAlb and CA3c showingthe blockageby CNQX (15 PM)and CPP(5 PM)of its initial part (middle traces). Note also the longerdelaybetweenthe CA1 and CA3LLDsunderthoseconditions.The effectsof theantagonists wereonly partially eliminatedby washing.B, SpontaneousLLDsrecordedin a CA3cneuron impaled with QX-3 14-filled microelectrodes to blockthe actionpotentialsassociated with theinitialbursts. Although CPP (2.5 NM)did not affect LLD shape, CNQX (3and9I.IM)blocked the initial burst.LLDs recordedunder thesecircumstances startedasan IPSP. Restingmembranepotentialof thecell in B was -55 mV.

the CA1 area along the Schaffer collaterals. In the adult rat hippocampus, 4-AP-induced epileptiform activity has previously beendescribedin CA3 (Voskyul and Albus, 1985; Rutecki et al., 1987; Perreault and Avoli, 1991). Concerning the CA1 region, two studieshave reported the occurrenceof spontaneous and evoked bursts (Voskyul and Albus, 1985; Holsheimer and Lopes da Silva, 1989) whereasthree others have describeda potentiation of synaptic transmission without any evoked or spontaneousbursts (Buckle and Haas, 1982; Segal, 1987; Perreault and Avoli, 1989). The resultsreported here indicate that some of these discrepanciesmight be due to technical factors. For example,when comparedto our previous findings(Perreault and Avoli, 1989) the modified slicing procedure in the present study resultedin the occurrenceof interictal events in the CA1 area. These observations emphasize,therefore, the importance of consideringthe physiological statusof the entire hippocampal slice even when investigations are carried out in one particular region. This is even more important for studieson experimental models of epilepsy in which the integrity of the CA3 area is pivotal in triggering epileptiform activity in other hippocampal areas(Wong and Traub, 1983). Field dischargesin CAl, as opposedto CA3, were not correlated intracellularly with bursts of action potentials. This is probably due to the fact that only a small number of CA1 neuronsfired synchronously during each event, asindicated by the relatively low amplitude of the population spikes on the CA1 field bursts. Hence, the probability of finding neuronsgenerating burstsof action potentials during eachepileptiform event remained relatively low. Despite the presenceof field potential discharges,single stimulation of the Schaffer collaterals failed in 80%of the slicesto evoke burstsin the CA1 region. However, a brief train of low-intensity stimulation was capable of producing a small epileptiform burst identical to spontaneously occurring interictal events. Presumably, the facilitation of GABAergic inhibition (Buckle and Haas, 1982;Perreault and Avoli,

112

Perreault

and

Avoli

* 4-AP

and

Rat Hippocampus

B Control

Control

4.,1 I

CA3-i

BMI

0 Ca++

Figure 9. Effects of 0 Ca2+ medium and various receptor antagonists on the occurrence of LLDs in slices where glutamatergic synapses were blocked. A, The top tracesshow low-speed chart records of LLDs occurring simultaneously in the CA1 and CA3 s. pyramidale of a slice where CNOX (20 UM) and CPP (5 PM) were also- present in addition to 4AP. When Ca2+ was removed from the medium (middle), LLDs were abolished and later reappeared after Caz+ reintroduction (bottom).B, The top tracesshow the LLDs recorded in CA1 and CA3 in a medium similar to the one in control traces in A. After the addition of BMI (50 PM), all activity disappeared (middle)and recovered after BMI washout (bottom).C, Graph shows effects of various receptor antagonists on the frequency of occurrence of LLDs. Only CY+ and BMI were effective in blocking the potential (p < 0.0 1). Calibration in B applies to A.

Wash

A

1

/

I

I

I

,

1

C

h .nnIL” 9 E 100

1989) combined with the lack of strong recurrent excitation (Knowles and Schwartzkroin, 1981; Christian and Dudek, 1988) prevents the generation of additional population spikesin CA 1 following single shock stimulations of the afferent fibers. 4-AP and GABA-mediated LLDs A lessusual type of spontaneous,synchronousactivity induced by 4-AP consisted of LLDs that occurred at a lower rate than the interictal dischargesand were generated simultaneously in all hippocampal areas. This type of activity has already been describedfor the CA1 region, where it is causedby the activation of GABA, receptors located on dendrites (Perreault and Avoli, 1989). The LLDs recorded in CA3 and dentate neurons were alsoblocked by the GABA, antagonistBMI and displayedproperties that were very similar to the LLDs recorded in CA1 . Our observations therefore indicate that, in all areas,GABA released periodically from intemeurons activates postsynaptic GABA, receptors that are presumably located in the dendrites, thereby producing a responsewhose hallmark is an LLD of both pyramidal and granule cells. The other GABA receptors appear also to be involved in the spontaneousLLDs, including somatic GABA, receptors, responsiblefor the initial hyperpolarization that usually precedesthe LLD, and GABA, receptorsthat generate the late hyperpolarizing phase of the potential (Perreault and Avoli, 1989). A similar potential has also been shown in 4-AP-treated slicesof the neostriatum (Kita et al., 1985) and, more recently, in slicesof the rat (Aram et al., 1991)and human neocortex (Avoli et al., 1988). The LLD describedhere is also

T

T

T

60 60 40

reminiscent of the type II spontaneousfield potential described by Voskyul and Albus (1985), who studied the effects of 4-AP in the hippocampal slice. One feature of the LLD was the occurrence in all regions of action potentials near its onset, usually on the peak of the early IPSP or on the initial part of the LLD’s rising phase.Evidence obtained in CA1 cells had shown that these action potentials were generatedectopically by the axons of pyramidal cells(Perreault and Avoli, 1989). Interestingly, excitatory amino acid antagonistshad no effect on the occurrence of theseaction potentials, indicating that they were not dependenton the function of glutamatergic synapses.One possibility is that the axonal regionsof pyramidal cellsare depolarized to threshold for action potential generation by a local rise in [K’], presumably due to the openingof repolarizing K+ channelsin the intemeurons that are firing during the LLD (seebelow). The blockage of axonal K+ current(s)by 4-AP could further raisethe level of excitability of the axon terminals (Kocsis, 1986). Simultaneousrecordingsin the slice’sthree main areasaswell as sectioning experiments indicate that the pacemaker of the LLD was not confined to a single region of the hippocampus but rather could be located in either the CAl, CA3, or dentate areas. Furthermore, once initiated in a given area, LLDs did not spread to the others in a constant or predictable pattern, suggestingthat many alternative propagation pathways could be utilized. Finally, the results obtained with the sectioning experiments demonstrate that the integrity of the fibers connecting the different areasof the hippocampusis not necessary

The Journal

for LLD propagation. For instance, LLDs remained coupled between CA 1 and dentate even following a complete sectioning of CA3 from the CAl-dentate regions. Although direct anatomical connections between CA 1 and dentate cells are absent (Lorente de No, 1934) GABAergic interneurons located in the s. lacunosum-moleculare in CA1 send both axonal and dendritic branches across the fissure, toward the dentate granule cell layer (Lacaille and Schwartzkroin, 1988a). The unique properties of these intemeurons could thus provide an anatomical substrate for the coupling of the LLDs in the two areas. However, the relatively slow propagation velocity observed (0.004-0.0 17 m/set) suggests that alternative nonsynaptic mechanisms are more likely to be involved. This is supported by the fact that LLDs remained coupled in all areas of the slice after the addition of both NMDA and non-NMDA antagonists that block fast synaptic transmission of many pathways in the hippocampus (Neumann et al., 1988; Davies and Collingridge, 1989; Lambert et al., 1989). Recent observations in the neocortex have also shown that the occurrence of GABA-mediated slow depolarizations is resistant to excitatory amino acid antagonists (Aram et al., 1991). It is likely, however, that in the intact slice, most of the LLD did propagate by making use of the well-known anatomical paths. For example, the propagation velocity of at least 34% of the LLDs traveling between CA 1 and CA3 was within the range reported for the Schaffer collateral fibers (Andersen et al., 1978; Holsheimer and Lopes da Silva, 1989). Also, current sourcedensity analysis of the CA1 LLD has indicated that spontaneous LLDs in this area are initiated by an input traveling along the Schaffer collaterals (Perreault and Avoli, 1989). Alternative pathways might be utilized only when these connections were anatomically or functionally disrupted during the various experimental conditions employed here. Accordingly, the mean propagation velocity of LLDs spreading between CA1 and CA3 decreased in a similar manner after section of the Schaffer collaterals or in the presence of excitatory amino acid antagonists. Synaptic transmission was required for the occurrence of LLDs because their appearance was abolished in 0 Caz+ medium or after the addition of Cd2+. These data, however, do not allow one to draw any conclusion concerning the propagation mechanism since the release of GABA from the intemeurons is presumably Ca*+ dependent as well. Moreover, the antagonists of a variety of transmitter receptors had no effect on the generation of LLDs, indicating that none of the considered putative transmitters were involved in the initiation and propagation of this activity. Although there are many other transmitter candidates that were not assessed in this study, the occurrence of LLDs was suppressed after the addition of BMI on slices where synaptic transmission at glutamatergic synapses had previously been blocked by the application of antagonists for both NMDA and non-NMDA receptors. It is likely, therefore, that the only synchronous activity that occurred under these circumstances was the GABA-mediated LLD generated in pyramidal or granule cells. LLDs recorded intracellularly during these experiments were initiated by a brief hyperpolarization, presumably an IPSP, which would confirm the fact that LLDs are not initiated by any excitatory input, unless this input impinges exclusively on the interneurons. It could also be that intemeurons are interconnected and excite one another by using GABA as a transmitter or other peptides that often colocalize with GABA in these cells such as somatostatin (Somogyi et al., 1984) enkephalin (Gall et al., 1981) or cholecystokinin (Greenwood et al.,

of Neuroscience,

January

1992,

12(l)

113

1981; Somogyi et al., 1984). However, electrophysiological evidence for direct interaction between intemeurons indicates that this is inhibitory (Misgeld and Frotscher, 1986; Lacaille et al., 1987; Lacaille and Schwartzkroin, 1988a,b), although this situation might be different when the amount of GABA released by inhibitory intemeurons is increased by 4-AP.

Mechanisms involved in the coupling of interneurons One of the main findings of this study is that GABAergic intemeurons located in different areas of the hippocampus interact by a mechanism that does not require excitatory synaptic transmission. The observations reported here might also suggest a simple model whereby LLDs are initiated by intemeurons located in any region of the hippocampus and propagate following the recruitment of other intemeurons located nearby and in other regions through the spatial dispersion of K+. According to this hypothesis, the firing of one interneuron will activate intrinsic repolarizing K+ conductances, thereby producing a local rise in [K+],, which will be, at least in part, dispersed by diffusion and by spatial buffering through the glial cell syncytium (Somjen, 1984). The elevated [K+], will depolarize neighboring intemeurons and enhance their excitability (Hounsgaard and Nicholson, 1983; Balestrino et al., 1986). The activity of these other cells will act as a positive feedback to promote further [K+], accumulation and dispersion and thus the firing of additional interneurons. This effect might be further facilitated by the lack of spike frequency adaptation in these cells (Schwartzkroin and Mathers, 1978; Lacaille and Schwartzkroin, 1988a,b). GABA-mediated potentials would be recorded in CA3 and CA1 pyramidal cells as well as in dentate granule cells as the wave of K+ spreads through the tissue. We also propose that an increase in intrinsic intemeuronal excitability (Segal, 1987) combined perhaps with a facilitation of GABA release at the synapse between the intemeurons and the pyramidal (Buckle and Haas, 1982) and granule cells might also contribute to LLD generation by further promoting the spillover of GABA on the extrasynaptic receptors where the depolarizing response is presumably generated (Alger and Ni~011, 1982). The increased firing rate of the intemeurons and the accumulation of [K+], would therefore be linked in a positive feedback loop (Hounsgaard and Nicholson, 1983) where 4-AP effects on the excitability of neurons will act as a reinforcing mechanism to promote their further discharge and the accumulation of [K+],. The K+ accumulation hypothesis that we are proposing is supported by preliminary data indicating that LLD occurrence is associated with prolonged (1 O-30 set) increases in [K+], of up to 6 mM (P. Perreault, C. Drapeau, and M. Avoli, unpublished observations). The role of K+ accumulation as a nonsynaptic propagation mechanism has also been suggested previously to explain the spread of seizure-like activity occurring in low-Caz+ solutions (Haas and Jefferys, 1984; Konnerth et al., 1986; Yaari et al., 1986). Interestingly, the range of propagation velocities reported for the low-Ca2+ bursts [O.OOl-O.l m/set (Haas and Jefferys, 1984) 0.2-0.5 mm/set (Konnerth et al., 1986)] was lower than the LLD propagation velocity recorded in intact slices, but closer to the velocities estimated between CA3 and CA1 when NMDA and non-NMDA receptors were blocked (0.02-0.2 m/set) or between dentate and CA1 areas when the CA3 area was sectioned (0.004-0.0 17 m/set). Specific interactions between neighboring neurons attributed to [K’], accumulation have also been

114

Perreault

and

Avoli

* 4-AP

and

Rat Hippocampus

reported in other preparations (Alkon and Grossman, 1978; Malenka et al., 198 1; Yarom and Spira, 1982). In principle, there are two additional nonsynaptic mechanisms, electrotonic coupling via gap junctions and ephaptic effects, that could facilitate the recruitment and the synchronization of firing of GABAergic interneurons locally. Electron microscopic observations have revealed the existence of gap junctions between various interneurons in the CAl, CA3, and dentate areas (Kosaka, 1983a,b); this type of coupling could therefore facilitate the svnchronization of activitv betweenthese cells. Electrical field (ephaptic) effects are, however, unlikely to play any significant role here becauseinterneurons are rather sparselydistributed in most regionsof the hippocampus(Ribak et al., 1978) and therefore are not closeenoughto one another to be sensitive to the extracellular field generatedby other intemeurons (cf. Haas and Jefferys, 1984; Taylor and Dudek, 1984).

What triggers the firing of the leading interneuron, and thus initiates the whole cycle in the hippocampal slice treated with 4-AP, is not clear. dne possibility is that4-AP might produce a membrane instability that allows interneurons to depolarize spontaneously.Segal(1987)hasindeed shownthat 4-AP applied focally in the CA 1areainducesrepetitive firing in intracellularly recorded interneurons and postulatedthat this activity could be responsiblefor the periodic appearancein pyramidal neurons of GABA-mediated hyperpolarizations. The propensity of the interneurons to undergo this kind of activity might therefore determine the frequency of occurrence of the LLDs reported in this study. References Alger BE (1984) Hippocampus: electrophysiological studies of epileptiform activity in vitro. In: Brain slices (Dingledine R, ed), pp 155199. New York: Plenum. Alger BE, Nicoll RA (1982) Pharmacological evidence for two kinds of GABA receptors on rat hippocampal cells studied in vitro. J Physiol (Lond) 328:125-141. Alkon DL, Grossman Y (1978) Evidence for non-synaptic neuronal interaction. J Neurophysiol 41:640-653. Andersen P, Silfvenius H, Sunberg SH, Sveen 0, Wingstrom H (1978) Functional characteristics of unmyelinated fibres in the hippocampal cortex. Brain Res 144:l l-18. Aram JA, Michaelson HB, Wong RKS (1991) Synchronized GABAeraic IPSPs recorded in the neocortex after blockade of svnantic transmission mediated by excitatory amino acids. J Neurophysioi65: 1034-1041. Avoli M (1990) Epileptiform discharges and a synchronous GABAergic potential induced by 4-aminopyridine in the rat immature hippocampus. Neurosci Let< 117:93-g%Avoli M. Perreault P. Olivier A. Villemure J-G (1988) 4-Aminoovridine induces a long-lasting depolarizing GABAergic potential inrhuman neocortical and hippocampal neurons maintained in vitro. Neurosci Lett 94:327-332. Balestrino M, Aitkin P, Somjen G (1986) The effects of moderate changes of K+ and Ca2+ on synaptic and neural function in the CA1 region of the rat hippocampal slice. Brain Res 377:229-239. Buckle PJ. Haas HL (1982) Enhancement of svnautic transmission bv 4-aminopyridine in hippocampal slices of the rat. J Physiol (Land) 326:109-122. Chestnut TJ, Swann JW (1988) Epileptiform activity induced by 4-aminopyridine in immature hippocampus. Epilepsy Res 2: 187195. Christian EP, Dudek FE (1988) Characteristics of local excitatory circuits studied with glutamate microapplication in the CA3 area of rat hippocampal slices. J Neurophysiol 59:90-109. Davies SN, Collingridge GL (1989) Role of excitatory amino acid receptors in synaptic transmission in area CA 1 of rat hippocampus. Proc R Sot Lond [Biol] 236:373-384.

Dutar P, Nicoll RA ( 1988) A physiological role for GABA, receptors

in the centralnervoussystem.Nature 332:156-l 58. Gall C, Brecha N, Karten HJ, Chang KJ (198 1) Localization of enkephalin-like immunoreactivity to identified axonal and neuronal populations of rat hippocampus. J Comp Neurol 198335-350. Greenwood RS, Godar SE, Reaves TA, Hayward JN (198 1) Chole-

cystokininin hippocampalpathways.J CompNeurol202:335-350.

Haas H, Jefferys JGR (1984) Low-calcium field bursts discharges of CA1 pyramidal neurones in rat hippocampal slices. J Physiol (Lond) 354:185-201. Holsheimer J, Lopes da Silva FH (1989) Propagation velocity of epileptiform activity in the hippocampus. Exu Brain Res 77:69-78. Honor& T, Davies- SN, Drejer J, Fletcher EJ, Jacobson P, Lodge D, Neilsen FE (1988)8) Quinoxalinediones: potent competitive nonNMDA glutamate receptor -----&-- antagonists. Science 241:701-703. Hounsgaard J, Nicholson C (1983) Potassium accumulation around individual Purkinje cells in cerebellar :erebellar slices ofthe guinea pig. J Physiol (Lond) 340:359-338.38. Kita T, Kita H, Kitai I ST (1985) Effects of 4-aminopyridine (4-AP) on rat neostriatal neurones rones in an in vitro slice preparation. Brain Res ?Al.lll-111

Knowles WD, Schwartzkroin PA (198 1) Local circuit synaptic interactions in hippocampal brain slice. J Neurosci 1:3 18-322. Kocsis J (1986) Functional characteristics of potassium channels of normal and pathological mammalian axons. In: Ion channels in neural membranes (Ritchie JM, Keynes RD, Bolis L, eds), pp 122-l 44. New York: Liss.

KonnerthA, HeinemannU, Yaari Y (1986) Nonsynapticepileptogenesis in the mammalian hippocampus in vitro. I. Development of seizurelike activity in low extracellular calcium. J Neurophysiol 56:

409-423. Kosaka T (1983a) Gap junctions between nonpyramidal cell dendrites in the rat hippocampus (CA 1 and CA3 regions). Brain Res 27 1: 15 7-

161.

Kosaka T (1983b) Neuronal gap junctions in the polymorph layer of the rat dentate gyrus. Brain Res 277:347-35 1. Lacaille JC, Schwartzkroin PA (1988a) Stratum lacunosum-moleculare interneurons ofhippocampal CA 1 region. I. Intracellular response characteristics, synaptic responses, and morphology. J Neurosci 8: 1400-1410. Lacaille JC, Schwartzkroin PA (1988b) Stratum lacunosum-moleculare intemeurons of hippocampal CA1 region. II. Intrasomatic and intradendritic recordings of local circuit synaptic interactions. J Neurosci 8:141 l-1424. Lacaille JC, Mueller AL, Kunkel DD, Schwartzkroin PA (1987) Local circuit interactions between oriens/alveus intemeurons and CA1 py-

ramidalcellsin hippocampalslices:electrophysiology andmorphology. J Neurosci 7: 1979-1993. Lambert JDC, Jones RSG, Andreasen M, Jensen MS, Heinemann U (1989) The role of excitatory amino acids in synaptic transmission in the hippocampus. Comp Biochem Physiol93A: 195-20 1. Lorente de No R (1934) Studies on the structure of the cerebral cortex. II. Continuation ofthe study ofthe ammonic system. J Psycho1 Neurol (Leipzig) 46: 113-l 77. Malenka RC, Kocsis JD, Ransom BR, Waxman SG (1981) Modulation of parallel fiber excitability by postsynaptically mediated changes in extracellular potassium. Science 2 14:339-34 1. Mihaly A, Bencsik K, Solymon T (1990) Naltrexone potentiates 4-aminopyridine seizures in the rat. J Neural Transm 79:59-67. Misgeld U, Frotscher M (1986) Post-synaptic-GABAergic inhibition of non-pyramidal neurons in the guinea pig hippocampus. Neuroscience 19: 193-206. Neumann RS, Ben-At? Y, Gho M, Cherubini E (1988) Blockade of excitatory synaptic transmission by 6-cyano-7-nitroquinoxaline-2,3 dione (CNQX) in the hippocampus in vitro. Neurosci Lett 92:64-68. Perreault P, Avoli M (1987) Effects of 4-AP on CA3 and dentate neurons of rat hippocampal slices. Sot Neurosci Abstr 13: 1160. Perreault P. Avoli M (1989) Effects of low concentrations of 4-aminopyridine of CA1 pyramidal cells of the hippocampus. J Neurophysiol 61:953-970. Perreault P, Avoli M (199 1) Physiology and pharmacology of epileptiform activity induced by 4-aminopyridine in rat hippocampal slices. J Neurophysiol 65:771-785. Ribak CE, Vaughn JE, Saito K (1978) Immunocytochemical localization of glutamic acid decarboxylase in neuronal somata following

The Journal

colchicine inhibition of axonal transport. Brain Res 140:3 1S-322. Rutecki PA, Lebeda FJ, Johnston D (1987) 4-Aminopyridine produces epileptiform activity in hippocampus and enhances synaptic excitation and inhibition. J Neurophysiol 57:1911-1924. Schwartzkroin PA, Mathers LH (1978) Physiological and morphological identification of a nonpyramidal hippocampal cell type. Brain Res 157:1-10. Segal M (1987) Repetitive inhibitory postsynaptic potential evoked by 4-aminopyridine in hippocampal neurons in vitro. Brain Res 4 14: 285-293. Somjen GG (1984) Interstitial ion concentration and the role of neuroglia in seizures. In: Electrophysiology of epilepsy (Schwartzkroin PA, Wheal H, eds), pp 304-341. London: Academic. Somogyi P, Hodgson AJ, Smith AD, Nunzi MG, Gario A, Wu JY (1984) Different populations of GABAergic neurons in the visual cortex and hippocampus of cat contain somatostatin- or cholecystokinin-immunoreactive material. J Neurosci 4:2590-2603. Spyker DA, Lynch C, Shabanowitz J, Jinn J (1980) Poisoning with 4-aminopyridine: report of three cases. Clin Toxic01 16:487-497.

of Neuroscience,

January

1992,

E’(1)

115

Szente M, Baranyi A (1987) Mechanism of aminopyridine-induced ictal seizure activitv in the cat neocortex. Brain Res 413:368-373. Taylor CP, Dudek FF (1984) Excitation of hippocampal pyramidal dells by an electrical field effect. J Neurophysiol 52: 126-142. Thesleff S (1980) Aminonvridines and svnantic _ - transmission. Neuroscience 5:1413-1419. - Voskyul RA, Albus H (1985) Spontaneous epileptiform discharges in hippocampal slices induced by 4-aminopyridine. Brain Res 342:5466. Wong RKS, Traub RD (1983) Synchronized burst discharge in disinhibited hippocampal slice. I. Initiation in CA2-CA3 region. J Neurophysiol49:442458. Yaari Y, Konnerth A, Heinemann U (1986) Non-synaptic epileptogenesis in the mammalian hippocampus in vitro. II. Role of extracellular potassium. J Neurophysiol 56:424-438. Yarom Y, Spira ME (1982) Extracellular potassium ions mediate specific neuronal interaction. Science 216:80-82.